概念：使用Python将字典结果的2D矩阵写入CSV文件

Question

I have dictionary formatted this way: the key is a tuple of the document number and the keyword, and the value is the frequency of the keyword within the document. So, the keys would be (document1, keyword1), (document1, keyword2), (document1, keyword3), (document2, keyword1), (document2, keyword2), (document2, keyword3), (document3, keyword1), (document3, keyword2), and (document3, keyword3) and the value would be number. Of course this is a small dictionary. I am hoping to apply the solution to a large set of documents and keywords.

The dictionary was created as such:

document_count = {}
try:
    for doc in document_id_list:
        indiv_doc = # records selected from a database
        for w in words:
            document_count.setdefault((doc, w), 0)
            for entry in #unsorted list of text tokenized, set to lower case, and stripped of stop words:
                if entry == w and (doc, entry) in document_count:
                        document_count[(patent, entry)] += 1
    return document_count

except Exception, e:
    print "create claim storages"
    print str(e)
    pass

I wanted to write the results to a CSV like a 2D matrix. At least, that is how I have seen it described.

      keyword1 keyword2 keyword3
document1 number   number   number
document2 number   number   number 
document3 number   number   number

When looking at the CSV function docs on python.org and other questions on this site, the closest that I have gotten to is this:

document1 keyword1 number
document1 keyword2 number
document1 keyword3 number
document2 keyword1 number
document2 keyword2 number
document2 keyword3 number
document3 keyword1 number
document3 keyword2 number
document3 keyword3 number 

This is the result from code that I have written:

 with open(os.path.join('C:/Users/Tara/PyCharmProjects/untitled/csv_results/', file_name),
                    'wb') as csvfile:
   w = csv.writer(csvfile)
   for key, value in available_dict.items():
       separate_keys = list(key)
       w.writerow([separate_keys[0], separate_keys[1], value])

I noticed that a lot of solutions involve list comprehension, but I do not know what the correct if statement would be. Would I make the changes when I write a dictionary, or when I write to the CSV file?

Answer 1

Many existing python libraries handle the task of writing a csv file, so I assume that you only want to use simple python statements and structures.

The main strategy below is to write a generator function to create the rows of the csv file. To do this, the function first extracts and sorts the documents and the keywords from the dictionary, then a header row is generated containing the keywords, then each document's row is created and generated

I'm using a minimal number of lists comprehensions, which could easily be avoided if you are ready to write a few more lines

D = {
    ('doc1', 'key1'): 2, ('doc1', 'key2'): 3, ('doc1', 'key3'): 4,
    ('doc2', 'key1'): 4, ('doc2', 'key2'): 6, ('doc2', 'key3'): 8,
    ('doc3', 'key1'): 6, ('doc3', 'key2'): 9, ('doc3', 'key3'): 12,
}

def gen_rows(D):
    sorted_docs = sorted(set(t[0] for t in D))
    sorted_kwds = sorted(set(t[1] for t in D))
    yield [None,] + sorted_kwds
    for d in sorted_docs:
        yield [d,] + [D.get((d, k), 0) for k in sorted_kwds]

for row in gen_rows(D):
    print(row)

Here is the output, a list of rows ready to be written in a csv file

[None, 'key1', 'key2', 'key3']
['doc1', 2, 3, 4]
['doc2', 4, 6, 8]
['doc3', 6, 9, 12]